Negative Returns: U.S. Military Policy and Anti-American Terrorism* Eugen Dimant†, Tim Krieger‡ and Daniel Meierrieks§

Economics Working Paper 17106

HOOVER INSTITUTION 434 GALVEZ MALL STANFORD UNIVERSITY STANFORD, CA 94305-6010

September 2017

We investigate the effect of U.S. military aid and U.S. troop deployments on anti-American terrorism, using a sample of 106 countries between 1986 and 2011. We find that greater military commitment leads to more anti-American terrorism. We study the underlying mechanisms using a mediation analysis and show that both U.S. military aid and troop deployments in foreign countries do not improve local state capacity. Rather, we find that more military aid (but not troop deployments) is linked to poorer political-institutional outcomes in aid-receiving countries, explaining the positive association between U.S. military aid and anti-American terrorism. Our findings suggest that U.S. military policy does not make the United States safer from transnational terrorism.

JEL Classification: D74; F35; F50 Keywords: U.S. military aid; U.S. troop deployments; anti-American terrorism; transnational terrorism; mediation analysis

The Hoover Institution Economics Working Paper Series allows authors to distribute research for discussion and comment among other researchers. Working papers reflect the views of the authors and not the views of the Hoover Institution.

* We benefited greatly from helpful suggestions and insightful discussions with Russell Berman, Guy Grossman, and Dan Hopkins. † Corresponding author. University of Pennsylvania. 249 S 36th Street, 19104 Philadelphia, USA. E-mail: [email protected]; and the Hoover Institution at Stanford University. ‡ University of Freiburg, Wilhelmsstraße 1b, 79098 Freiburg, Germany. E-mail: [email protected]; and CESifo, Munich. § University of Freiburg, Wilhelmsstraße 1b, 79098 Freiburg, Germany. E-mail: [email protected].

1 1. Introduction

In 2016 the U.S. spent approximately $6 billion on foreign military aid (U.S. Department of

State, 2017). According to Vine (2015), the costs of operating American military bases overseas

(outside warzones) amounted to $85 to $100 billion in the fiscal year 2014 alone. Besides the promotion of stability in aid-receiving countries, the safeguarding of U.S. homeland security from foreign terrorist threats is among the main goals of these military policy measures (The

White House, 2013). Interestingly, however, the small number of empirical studies that have examined the determinants of transnational anti-American terrorism1 provide evidence that more military assistance by the United States is associated with more anti-American terrorism originating from the assistance-receiving country (Neumayer and Plümper, 2011; Gries et al.,

2015; Krieger and Meierrieks, 2015; Saiya et al., 2017). This raises doubts as to whether the enormous resources used for U.S. military policy measures are well spent.

The aforementioned studies, however, do not examine which underlying mechanisms can explain the nexus between U.S. military assistance and anti-American terrorism. We fill this research gap by empirically studying this link by means of a mediation analysis. A preview of our findings indicates that U.S. military assistance is unsuccessful in contributing to local state capacity building and thus unsuccessful in inhibiting anti-American terrorism. Rather, higher levels of U.S. military assistance (in the form of financial military aid) are associated with poorer political-institutional outcomes (e.g., more corruption, poorer human rights records) and more anti-American terrorism, suggesting that the U.S. is being punished in the form of anti-

American terrorism for the—ostensible or actual—facilitation of local grievances.

1 Transnational terrorism means that more than one country is affected by a terrorist incident. For instance, the 9/11 attacks were transnational because foreign terrorists attacked on U.S. soil, with non-Americans being also victimized.

2 We proceed as follows. In Section 2 we discuss the theoretical mechanisms explaining how

U.S. military policy may translate into anti-American terrorism. We introduce our methodology and data in Section 3. Section 4 presents our empirical findings. Section 5 concludes.

2. Transmission Channels

Two potential mechanisms may explain the correlation between U.S. military policy and the genesis of anti-American terrorism, the state capacity channel and the grievances channel.

State Capacity Channel. Anti-American terrorism may be the consequence of conflict between a local terrorist group and a U.S.-backed local government (Addison and Murshed, 2005). U.S. military sponsorship ought to increase the local government’s state capacity by, for example, freeing up government resources to improve the provision of public goods and appease the opposition (Addison and Murshed, 2005). Strengthening the aid-receiving government, however, also creates a strategic incentive for the opposing terrorist organization to attack the

United States: Anti-American terrorism may lead the U.S. to withdraw its support for the local government (e.g., due to political pressure from the U.S. Congress or media), thus increasing the likelihood of terrorist success by weakening the opposing local government (Plümper and

Neumayer, 2011). In other words, by strengthening the local government, the United States, paradoxically, may become a likelier target of transnational terrorism by violent domestic challengers of the government in place.

Grievances Channel. Anti-American terrorism may also result from the—ostensible or actual—negative repercussions of U.S. military assistance in aid-receiving countries. As for actual repercussions, U.S. military assistance may contribute to poorer political-institutional outcomes in an aid-receiving country (as indicated by, e.g., corruption and human rights violations). For instance, Djankov et al. (2008) argue that foreign aid can be considered a

3 windfall gain (i.e., an unexpected increase in national income) that fuels rent-seeking behavior and corruption and reduces incentives for political-institutional reforms. When U.S. military assistance perpetuates or further erodes unfavorable political-institutional conditions in aid- receiving countries, this can consequently be expected to give rise to anti-Americanism. In addition to its actual consequences, the perception of military assistance by the local population in the aid-receiving country may also matter. For instance, Gries et al. (2015: 87) argue that by providing a repressive regime with aid, the United States “becomes associated with [...] local repression,” even though this aid most likely does not directly finance local authoritarianism.2

Generally speaking, U.S. support for an unpopular local government (plagued by, e.g., corruption and human rights violations) may mean that resentment is projected onto the United

States and results in anti-American terrorism. For instance, in his 2002 “Letter to America”

Osama bin Laden (2002) laments corruption, repression and economic mismanagement in the

Islamic world but blames these factors on U.S. involvement, justifying Al-Qaeda’s anti-

American terrorist activities.

3. Methodology and Data

3.1. Methodology

To investigate the relationship between U.S. military policy and anti-American terrorism and the roles state capacity and local grievances may play in mediating this relationship, we follow the causal mediation framework described in, e.g., Imai et al. (2010, 2011). Consider the following set of regressions that include variables for anti-American terrorism (terror), U.S.

2 In fact, U.S. law (e.g., the 1997 Leahy amendment) explicitly curtails the provision of military assistance when human rights are violated.

4 military policy (uspolicy), factors mediating the relationship between the two (mediator) as well as a vector of confounding controls (X):

푇 terror푖푡 = ∝1 + 훽1푢푠푝표푙푖푐푦푗,푖푡 + 휁1 푋푖푡 + 휀푖푡1 (1)

푇 mediator푗,푖푡 = ∝2 + 훽2푢푠푝표푙푖푐푦푗,푖푡 + 휁2 푋푖푡 + 휀푖푡2 (2)

푇 terror푖푡 = ∝3 + 훽3푢푠푝표푙푖푐푦푗,푖푡 + 훾mediator푗,푖푡 + 휁3 푋푖푡 + 휀푖푡3 (3)

Equation (1) examines whether measures of U.S. military policy affect anti-American terrorism. Due to the nature of the dependent variable (described below in more detail), we run a series of probit maximum-likelihood estimations to estimate the effect of U.S. military policy on anti-American terrorism. To ameliorate endogeneity concerns3 we also report findings from instrumental-variable probit estimations, where we use the ten-year lag of our military policy variables as our instrumental variables.

While equation (1) mainly serves to corroborate previous research on the effects of U.S. military policy on anti-American terrorism, equations (2) and (3) are crucial to identifying mediation effects. Here, we first fit models for the mediator variable and the outcome. For parametric inference, we simulate model parameters from their sampling distribution and repeat the simulation of the potential values of the mediator and of the potential outcome given the simulated values of the mediator for an appropriate number of times (1,000 simulations). This allows us to compute the causal mediation effects, before computing summary statistics like point estimates and confidence intervals (Imai et al., 2010: 317). Importantly, this approach allows us to accommodate nonlinear relationships given that we consider a probit model,

3 For instance, endogeneity may be due to simultaneity, where U.S. military policy also responds to anti-American terrorism (Boutton and Carter, 2014).

5 conduct sensitivity analysis4 and provide an estimate of the average mediation effect as a percentage of the total effect to quantify the strength and direction of the mediation (Imai et al.

2010, 2011).

3.2 Data

For our analysis we use a balanced panel of 106 countries for the time period of 1986-2011.

The summary statistics, data sources and operationalization of all variables discussed below are reported in Table 1. A country list is provided in the appendix.

Terrorism. Anti-American terrorism is measured by a binary variable that takes on the value 0 when there is no anti-American terrorism and the value 1 when there is at least one terrorist attack by a citizen of another country against U.S. targets in a given country per year. Such targets may include U.S. embassies, tourists, companies and military personnel, excluding attacks against U.S. military targets associated with declared wars, military interventions or guerilla warfare (e.g., insurgent attacks against U.S. military targets during the U.S. intervention in Iraq). The data are constructed from the International Terrorism: Attributes of Terrorist

Events (ITERATE) dataset of Mickolus et al. (2015).

4 The causal mediation approach relies on the sequential ignorability assumption (see Imai et al., 2011: 770-772 for more details). If this assumption is violated, the mediated effect is potentially sensitive to unmeasured confounders (Imai et al., 2011: 776). A violation of this assumption leads to a correlation between 휀푖푡2 and 휀푖푡3 in equations (2) and (3), where this correlation is denoted as ρ (with ρ=0 implying that the sequential ignorability assumption holds). For sensitivity analysis purposes, we report the value of ρ for which the calculated average causal mediation effect would be equal to zero, with larger values of ρ coinciding with more “robust” findings. Note, however, that there is no known scale for evaluating what qualifies as small, medium or large correlations between residuals.

6 Variable Observations Mean Std. Dev. Min. Max. Variable Definition Source Anti-American Terrorist Attacks 2,756 0.13 0.33 0 1 Population Size 2,756 7.08 1.55 3.19 11.81 Total size of population, logged (a) Per Capita Income 2,756 8.74 1.25 4.89 11.96 Real GDP per capita at constant national prices (in mil. (a) 2011US$), logged Democracy 2,733 6.71 3.42 0 10 Unified polity score, ranging from 0 (full autocracy) to +10 (b) (full democracy) Distance to U.S. 2,756 5.26 2.20 0.01 13.36 Distance between Washington, DC, and the respective foreign (c) country’s capital, in 1,000km Civil War 2,756 0.05 0.22 0 1 Dummy variable, takes on value 1 if country experiences more (d) than 1,000 battle deaths in a specific year CINC Score 2,756 0.62 1.73 <0.01 21.22 Composite measure of national power accounting demographic (e) (e.g., urbanization), economic (e.g., energy consumption) and military strength (e.g., military expenditure) Relative Political Extraction 2,722 0.97 0.35 0.07 2.70 Approximates the ability of governments to appropriate (f) portions of the national output to advance public goals (by accounting for tax revenue, agricultural income, GDP etc.) Rule of Law 2,756 7.03 2.50 1 11 Assessment of the strength and impartiality of the legal system (g) and popular observance of the law Physical Integrity Rights 2,656 4.71 2.32 0 8 Additive index indicating presence of torture, extrajudicial (h) killings, political imprisonment and disappearances Women Political Rights 2,658 1.89 0.61 0 3 Indicates extent of political discrimination of women (e.g., (h) right to vote and run for office) within a society Corruption 2,756 5.95 2.25 1 11 Assessment of corruption related to patronage, nepotism, (g) suspiciously close ties between politics and business etc. Marketization 2,513 3.69 2.01 0.05 8.89 Per capita life insurance premiums in force (in 2005 (i) international $), logged (as a measure of contract-intensity) Data Sources: (a) Penn World Tables 9.0 (http://www.rug.nl/ggdc/productivity/pwt/); (b) Polity IV Project (http://www.systemicpeace.org/inscrdata.html); (c) CEPII GEODist Database (http://www.cepii.fr/cepii/en/bdd_modele/presentation.asp?id=6); (d) UCDP/PRIO Armed Conflict Dataset (https://www.prio.org/Data/Armed- Conflict/UCDP-PRIO/); (e) National Material Capabilities 5.0 (http://cow.dss.ucdavis.edu/data-sets/national-material-capabilities/national-material-capabilities-v4- 0); (f) Relative Political Capacity Dataset (https://dataverse.harvard.edu/dataset.xhtml?persistentId=hdl:1902.1/16845); (g) International Country Risk Guide (http://epub.prsgroup.com/products/international-country-risk-guide-icrg; (h) CIRI Human Rights Dataset (http://www.humanrightsdata.com/); (i) CINE Dataset (https://sciences.ucf.edu/politicalscience/people/mousseau-michael/). Table 1: Summary Statistics and Variable Operationalization

7 US. Military Policy. We employ two measures of U.S. military policy. First, we use data on

(logged) financial military aid in constant 2014 U.S.-$ from USAID (2015). USAID (2015) defines military assistance as aid for programs primarily for the benefit of a recipient government’s armed forces or aid, which subsidizes local military capability. Second, we use data on U.S. troop deployments on foreign soil, using data from Kane (2016).9

Mediators. In Section 2, we discussed two potential channels from U.S. aid to anti-American terrorism: the state capacity channel and the grievances channel. We use the following set of variables to capture these channels. We use three measures of state capacity, one capturing a country’s strength in terms of economic-industrial, military and demographic power (CINC score), one related to the ability of a government to utilize economic output to promote public goals, e.g., with respect to taxation (relative political extraction) and the third indicating the strength of a country’s police and judicial system (rule of law). In line with our discussion of the state capacity channel, U.S. military policy is expected to promote local state strength, which in turn is expected to incentivize terrorism by local terrorist groups against the United

States. Four additional variables indicate local grievances, measuring the degree of human rights violations (physical integrity rights), the role of women in society (women political rights), corruption and the degree to which local economic transactions are carried out through markets rather than clientalism (marketization). In line with the grievances channel, we hypothesize U.S. military policy to be associated with poorer political-institutional outcomes

(i.e., less respect for human and women’s rights, more corruption and smaller emphasis on markets), which in turn ought to result in more anti-American terrorism.

9 Dropping outliers in terms of U.S. financial aid (e.g., Israel) and U.S. troop deployments (e.g., Iraq) does not change the main results reported below. Also, our results do not change when we express the U.S. military policy variables in per capita terms rather than in levels. These results are available upon request.

8 Controls. For the controls, we follow previous studies on the determinants of anti-American terrorism (Neumayer and Plümper, 2011; Gries et al., 2015; Krieger and Meierrieks, 2015;

Saiya et al., 2017) and control for local conditions in countries producing anti-American terrorism: population size, per capita income, democracy, geographical distance to the U.S. and civil war. Year and continent dummies are also included in all estimations.

4. Empirical Results

First, we estimate equation (1) to corroborate the findings from previous empirical studies.

Indeed, as shown in Table 2, both probit and IV-probit models indicate that more U.S. military aid as well as larger troop deployments are associated with more anti-American terrorism, mirroring previous empirical efforts that have examined the role of U.S. military policy in the emergence of anti-American terrorism (Neumayer and Plümper, 2011; Gries et al., 2015;

Krieger and Meierrieks, 2015; Saiya et al., 2017).10 The results for the controls are as expected.

10 Our findings are robust to the inclusion of additional controls including, e.g., trade openness, oil wealth, urbanization, ethnic fractionalization and population density as well as to restricting our analysis to the post-Cold War era (1994-2011) and the post-9/11 era (2002-2011). Using these robustness checks and sub-samples, we arrive at the same results as reported in the main text. Results are available upon request.

9 (1) (2) (3) (4) U.S. Military Aid 0.024 0.045 (0.010)** (0.020)** U.S. Troop Deployment 0.134 0.168 (0.034)*** (0.048)*** Population Size 0.198 0.193 0.158 0.142 (0.048)*** (0.047)*** (0.049)*** (0.052)*** Per Capita Income 0.031 0.056 -0.137 -0.172 (0.076) (0.077) (0.079)* (0.089)* Democracy -0.024 -0.033 -0.022 -0.023 (0.025) (0.025) (0.024) (0.023) Distance to U.S. -0.084 -0.086 -0.042 -0.033 (0.043)* (0.042)** (0.049) (0.050) Civil War 0.753 0.781 0.748 0.759 (0.193)*** (0.197)*** (0.193)*** (0.194)*** Year Dummies Yes Yes Yes Yes Continent Dummies Yes Yes Yes Yes Empirical Method Probit IV-Probit Probit IV-Probit Wald χ2 Test of Exogeneity 2.22 1.68 (Prob > χ2) (0.14) (0.20) No. of Observations 2,627 2,627 2,627 2,627 Notes: Dependent variable: Number of anti-American terrorist attacks. Constant not reported. Probit and IV-Probit regression results reported. Instruments for IV-Probit: U.S. military aid t-10 and U.S. troop deployment t-10. Country-clustered standard errors in parentheses. *p<0.1, **p<0.05, ***p<0.01. Table 2: U.S. Military Policy and Anti-American Terrorism

To understand why a stronger military commitment by the U.S. results in more anti-American terrorism, we run the mediation analysis as described above. Its results are reported in Table 3.

State Capacity Channel. Our mediation analysis provides no support for the state capacity channel. For one, we detect no effect of the military policy variables on relative political extraction and the CINC score. That is, U.S. military policy is unsuccessful in contributing to local state capacity building, which is in sharp contrast to the goals outlined by various U.S. administrations (e.g., The White House, 2013).

For another, focusing on the local rule of law, we find that more U.S. military aid actually reduces rather than strengthens it. The erosion of the rule of law, in turn, contributes substantially to the overall effect of U.S. military aid on anti-American terrorism. This finding may suggest that U.S. aid facilitates the use of unlawful means by an aid-receiving government such as the use of paramilitaries (Dube and Naidu, 2015). An alternative interpretation of this

10 finding is related to the perception of military aid. Here, U.S. assistance for a local government that disregards the rule of law may lead to local dissatisfaction that consequently spills over into anti-American terrorism. This latter interpretation of our finding already relates to the grievances—ostensible or actual—U.S. military policy measures may generate.

Grievances Channel. Studying the grievances channel in more detail, we find further support for its relevance in explaining anti-American terrorism. Consistent with our expectations, we find that poor political-institutional conditions are associated with more anti-American terrorism. For instance, these results are in line with Krieger and Meierrieks (2015) in that low levels of marketization lead to anti-American terrorism and with Saiya et al. (2017) in that women’s rights violations are associated with more anti-American terrorist attacks. More importantly, more U.S. military aid is associated with a deterioration of these very political- institutional conditions, contributing substantially to the overall positive effect of U.S. military aid on anti-American terrorism. For one, this finding may suggest that the U.S. is being punished when U.S. military aid facilitates local grievances. For instance, financial military aid may fuel local corruption or allow local regimes to postpone institutional reforms. For another, the perception of aid may also play a role. When the U.S. provides aid to an unpopular local government (because, e.g., corruption and clientalism are rampant), some of the local discontent may also be projected onto the United States, leading to anti-American terrorism.

Military Aid vs. Troop Deployments. Notably, our mediation analysis shows that U.S. military aid has a much stronger effect on local institutional conditions than U.S. troop deployments.

Arguably, financial aid is much more likely to be misused by local governments to perpetuate poor local institutional conditions, for example, using financial aid to cross-finance repression or buying off political opposition through corruption. U.S. troops, by contrast, are under direct

U.S. control, making their misuses by local governments—and thus their effect on local institutions—much less likely.

11 Mediator (M) Aid  M M  Percent ρ at which Terrorism Mediated ACME=0 CINC Score -0.009 -0.079 2.93% -0.1 (0.019) (0.039)** Relative Political Extraction 0.003 -0.248 -3.42% -0.1 (0.003) (0.169) Rule of Law -0.042 -0.134 23.67% -0.2 (0.014)*** (0.036)*** Physical Integrity Rights -0.034 -0.181 24.75% -0.2 (0.016)** (0.035)*** Women Political Rights -0.008 -0.307 10.15% -0.1 (0.005)* (0.089)*** Corruption 0.043 0.121 20.62% 0.1 (0.015)*** (0.034)*** Marketization -0.042 -0.103 19.80% -0.1 (0.010)*** (0.061)* Mediator Troops  M  Percent ρ at which M Terrorism Mediated ACME=0 CINC Score -0.070 -0.068 3.51% -0.1 (0.067) (0.045) Relative Political Extraction -0.009 -0.188 1.28% -0.1 (0.014) (0.183) Rule of Law 0.018 -0.155 -2.18% -0.2 (0.046) (0.038)*** Physical Integrity Rights 0.036 -0.214 -5.39% -0.2 (0.052) (0.033)*** Women Political Rights -0.048 -0.263 9.22% -0.1 (0.021)** (0.098)*** Corruption 0.079 0.112 6.91% 0.1 (0.051) (0.034)*** Marketization 0.002 -0.154 -0.07% -0.1 (0.054) (0.062)** Notes: For all models and tests the following covariates are included: population size, income p.c., democracy, distance to the United States, civil war, time dummies and continent dummies. 1000 simulations run for the quasi-Bayesian approximation of parameter uncertainty. ACME=Average causal mediation effect. Country-clustered standard errors in parentheses. *p<0.1, **p<0.05, ***p<0.01. Table 3: Mediation Analysis

5. Conclusion

We study the effect of U.S. military policy on anti-American terrorism for a sample of 106 countries from 1986 to 2011. We find that more active U.S. military policy translates into more anti-American terrorism produced in the target country of these very policies, corroborating previous evidence on the determinants of anti-American terrorism. We show that this effect is not due to U.S. policies strengthening local state capacity and thus creating a strategic incentive for local terrorist groups to engage in anti-American terrorism. Rather, we find that more U.S.

12 military aid (but not U.S. troop deployments) is associated with poorer political-institutional conditions, which gives rise to grievances and anti-American terrorism in aid-receiving countries. That is not to say that all foreign aid will always yield such negative returns. In light of our empirical findings, military measures (especially military aid), however, do not appear to be an appropriate policy tool to better protect the United States from transnational terrorism.

References

Addison, T., Murshed, S.M. (2005). Transnational terrorism as a spillover of domestic disputes

in other countries. Defence and Peace Economics 16, 69-82.

Bin Laden, O. (2002). Letter to America. Available at

https://www.theguardian.com/world/2002/nov/24/theobserver.

Boutton, A., Carter, D.B. (2014). Fair-weather allies? Terrorism and the allocation of US

foreign aid. Journal of Conflict Resolution 58, 1144-1173.

Djankov, S., Montalvo, J.G., Reynal-Querol, M. (2008). The curse of aid. Journal of Economic

Growth 13, 169-194.

Dube, O., Naidu, S. (2015). Bases, bullets, and ballots: The effect of US military aid on political

conflict in Colombia. Journal of Politics 77, 249-267.

Gries, T., Redlin, M., Meierrieks, D. (2015). Oppressive governments, dependence on the USA,

and anti-American terrorism. Oxford Economic Papers 67, 83-103.

Imai, K., Keele, L., Tingley, D. (2010). A general approach to causal mediation analysis.

Psychological Methods 15, 309-334.

Imai, K., Keele, L., Tingley, D., Yamamoto, T. (2011). Unpacking the black box of causality:

Learning about causal mechanisms from experimental and observational studies. American

Political Science Review 105, 765-789.

Kane, T. (2016). The Decline of American Engagement: Patterns in U.S. Troop Deployments.

Hoover Institution Economics Working Paper 16101.

13 Krieger, T., Meierrieks, D. (2015). The rise of capitalism and the roots of anti-American

terrorism. Journal of Peace Research 52, 46-61.

Mickolus, E.F., Sandler, T., Murdoch, J.M., Flemming, P.A. (2015). International Terrorism:

Attributes of Terrorist Events, 1968-2014. Vinyard Software: Dunn Loring, VA.

Neumayer, E., Plümper, T. (2011). Foreign terror on Americans. Journal of Peace Research 48,

3-17.

Saiya, N., Zaihra, T., Fidler, J. (2017). Testing the Hillary doctrine: Women’s rights and anti-

American terrorism. Political Science Quarterly 70, 421-432.

The White House (2013). Fact Sheet: U.S. Security Sector Assistance Policy. Available at

https://fas.org/irp/offdocs/ppd/ssa.pdf.

USAID (2015). U.S. Overseas Loans and Grants Greenbook. Available at

https://explorer.usaid.gov/reports.html.

U.S. Department of State (2017). Congressional Budget Justification, Fiscal Year 2017.

Available at https://2009-2017.state.gov/s/d/rm/rls/ebs/2017/pdf/index.htm.

Vine, D. (2015). Base Nation: How U.S. Military Bases Abroad Harm America and the World.

Metropolitan Books, New York.

Appendix. List of Countries

Albania, Algeria, Angola, Argentina, Australia, Austria, Bahrain, Bangladesh, Belgium,

Bolivia, Botswana, Brazil, Bulgaria, Burkina Faso, Cameroon, Canada, Chile, China,

Colombia, Congo, Congo (DR), Costa Rica, Cyprus, Denmark, Dominican Republic, Ecuador,

Egypt, El Salvador, Ethiopia, Finland, France, Gabon, Gambia, Ghana, Greece, Guatemala,

Guinea, Guinea-Bissau, Haiti, Honduras, Hungary, Iceland, India, Indonesia, Iran, Iraq, Ireland,

Israel, Italy, Jamaica, Japan, Jordan, Kenya, Kuwait, Lebanon, Liberia, Madagascar, Malawi,

Malaysia, Mali, Malta, Mexico, Mongolia, Morocco, Mozambique, Netherlands, New Zealand,

Nicaragua, Niger, Nigeria, Norway, Oman, Pakistan, Panama, Paraguay, Peru, Philippines,

14 Poland, Portugal, Qatar, Romania, Saudi Arabia, Senegal, Sierra Leone, Singapore, South

Africa, Spain, Sri Lanka, Sudan, Suriname, Sweden, Switzerland, Syria, Tanzania, Thailand,

Togo, Trinidad and Tobago, Tunisia, Turkey, Uganda, United Kingdom, Uruguay, Venezuela,

Viet Nam, Zambia, Zimbabwe.

15